all2

I don't want to hijack your thread. My email is in my profile, I'd be glad to shoot you an in depth description of what I have so far.

I'm currently working on an agent framework that has auditability as one of its core promises. I'm glad to see others are working in this domain!

I've seen other products/apps in this space farther up the stack at the API boundary.

What frameworks does your package work with? How does it handle intercept?

Money. Someone made a lot of money selling IP and the skills of the team.

Qwen 3.6 27b at q6 works decently well on 24GB of VRAM

It's an open source project. Maybe he'll accept PRs?

Time-cost for machines instead of willing knowledgeable humans. The former requires money, the latter requires passion.

Arguably, passion for a project is without price.

The WRT54 is still one of my favorite pieces of industrial design. Small case, ~~purple~~blue and grey, two antennas.

https://en.wikipedia.org/wiki/Linksys_WRT54G_series

This [0] may be a good starting point. The crux is EMFs can interfere with 'voltage gated calcium channels' in cell walls.

[0] https://www.sciencedirect.com/science/article/pii/S001393511...

When my brother was in town one Christmas break we logged ~100 hours over 6 days playing coop.

This happens to me occasionally, but it's typically stimulant and blood sugar related.

There are some indications that it might be related to ion availability in your body, so copper, zinc, and calcium specifically. You may also consider that HF RF found in phones, wifi, and Bluetooth can do not great things to the calcium channels in cell walls (it essentially locks the channels open in some cases).

There are a lot of factors to this, but these are things I've picked up in my reading for my own issues.

Perfect. I love it.

*pique

'peak' refers to the top of a thing, commonly mountains

LLMs are trained on the Internet, which isn't exactly known for it's race agnostic opinions.

Don't buy new. Get a 10 year old dual processor server tower on eBay with 128GB of RAM and a 1TB SSD.

It's cheaper than new and it'll do everything you need it to without even breaking a sweat.

Is think you would use it more like email and less like text messages, so the domain of communication shifts drastically. The other part is, you don't have to run just that model, you can offload a lot of chores to smaller models.

I know. It hurts to let things die in my project backlog. But there's so much of 'life' outside the project log that I don't have time. I have to prioritize.

I feel the 'don't waste time on HN' thing. I'm working on it, minimizing social media usage, minimizing non-productive screen time.

The urge to tear down the stack of cellphones I have and pull the boot flash chip hits me occasionally. It would be a substantial project, though, so I haven't done it. Yet.

In a general sense, you see this crop up in SKILLS.md files and other places, as LLMs have to deal with broad contexts. People try and drill into some taxonomy in a naive fashion using plaintext as a directive, which is not particularly optimal.

I wonder if one could build a 'mixture of experts' at the model level that leveraged a variety of small models "within" a larger model...

These are absurdly effective for this kind of task. Training is fast and straight forward. Packaging for deployment as ONNX is pretty simple as well.

Got a PDF? I love old books like this.

This sounds amazing.

I may have to dive back in. I've been using lazyvim for most of my terminal work now because spacemacs was too much effort to keep running, where-as lazyvim is a git pull and done sort of thing.

I think so, yes. I didn't hear great things about that. Eventually I'll probably watch it. But also maybe not.

The danger of "more" is that it dilutes the purpose and voice of the original. "Cowboy Bebop" fits in this same realm, I think. It had a single season. They did a movie. They said everything they needed to say and left it at that.

I have so much nostalgia for Watterson's work. I occasionally will buy another of the hard bound 3 volume set. I always wind up giving them away and then buying another.

A worthy cause, I hope.

They are, its the exchange rate that's the killer.

There are services where you can hook your card up and rent it out to other users. I don't know what any of them are called, but they do exist.

Ensemble or 'mixture of experts' (before MoE model architecture was a thing).

> Do you review all the files that are generated to ensure there's no hallucinations? Do you just review the last file of concise findings instead?

Sometimes, yes. Typically I'll read the final fusion doc, and then trace backwards if there is something that looks relevant. Sometimes I'll read all the abstracts as they come through.

> Is the intent here that the hallucinations will be countered by running through multiple agents that you end up with only the truth? Have you seen anything in the last version that was egregiously wrong?

The intent isn't so much avoiding hallucinations, rather I was attempting to acquire unique, domain specific insight. My read-through of the final doc is a 'pick and choose', where I weed out what isn't relevant, and keep what is. I haven't seen anything way out of whack, as agents will typically check each other and call each other out in the review/rebuttal stage.

> So this is all just them running with the context of the file, without reference tot he rest of the project?

This is run in an agent harness locally, so each reviewer has access to the whole project. The review rebuttal stage rarely sees agents re-read files, though, unless one of them is particularly aggressive and is going deep on something (GPT 5 series will do this to make a point, sometimes).

> Thanks for any responses to this.

No worries. This is pretty easy to try, even with something you don't own. You can run it in any harness that allows spawning sub-agents.

If there's a lot of interest, I could spin up a simple web app that does this just so folks can see it churn on a target git repo or project files or whatever.

The idea is pretty simple, identity drives focus and outcomes of each of the 'agents'. The review/rebuttal stage filters stuff that likely doesn't matter in the grand scheme (you'll find reviewers get pedantic and take hard lines on things that ultimately don't matter very much).

This is based on earlier work from late '25 where people were doing similar things. My added bit is the 'unique identities'. What I found was that the root agent typically picks identities relevant to the project, and so you get relevant 'data' or 'views' from a variety of angles.